Hubsan X4 Replacement Frame

Project Logs

Collapse

Extension Wires, First Flights, and Learning from Failure
zakqwy • 03/27/2015 at 14:22 • 6 comments

Nothing like a positive title, right? Spoiler: the first flight didn't go exactly as planned, but with any luck I learned a few lessons.
My 28 AWG wire assortment and liquid electrical tape (yup, this is a thing!) arrived earlier this week. I reinstalled the four motors into their mounts, threaded the wires through the arms, and soldered on appropriately colored extensions:
Lesson 1: don't assume you can estimate wire gauge when you're working on a scale you aren't used to. I totally could have gotten away with 32 AWG; the 28 AWG stock, partially due to its rather thick insulation, isn't terribly flexible. Oh well, time for some liquid electrical tape:
Now this stuff is cool. Doesn't smell great, but it's definitely superior to heat shrink tubing for tiny projects like this. After the insulation dried in a few minutes, I soldered half the board on:
Plenty of space for soldering. Next, the remaining four motor terminals get soldered:
Lesson 2: just because wires were too short before doesn't mean you should use waaaay too much wire now. Once I got to this point I considered chopping the eight extensions back a bit, but I decided to bolt everything together to run an initial flight test:
Oh man, totally not good! Okay, so what happened?
I initially paired the transmitter to the X4 and strapped the battery on with a cut down strip of tie wrap. I pressed on one propeller and slowly advanced the throttle to verify correct propeller pitch and motor polarity; after getting this right, I mounted up the other three props and verified operation of all four motors.
Lesson 3: Just because the props spin doesn't mean it's ready to fly. Having verified propeller orientation and motor operation, I blasted the throttle to 100% with the intention of flying a quick victory lap around the workshop. Upon leaving my desk the X4 immediately started yawing at a blistering rate--probably 8-10 RPM. I pulled the throttle back and tried for a heroic catch, but the quad hit the table and snapped off a leg. Undeterred, I clipped off the other three legs (gotta maintain symmetry!) and prepared for another try.
Okay, to be fair--at this point, I did try a few things. I triple-checked the motor and prop installation and verified that everything was rotating in the proper direction. I also recalibrated the X4's accelerometer per the Hubsan manual for good measure. Then I jammed the throttle forward again and watched helplessly as the craft once again spun to its doom; this time, I didn't pull back as quickly and the craft clattered to the floor on the other side of the room. The resulting damage shown above meant that this was the last test of the evening.
What other lessons have I learned?
Lesson 4: Quads are highly engineered and tuned systems. While I'm confident that my design is fairly balanced, I believe that the uncontrolled yawing was caused by angular errors in the motor orientation since I didn't constrain the mounts during glue-up:
Maybe that's why most (okay, pretty much all) microquads constrain motor orientation precisely, either using a mass-produced housing (i.e. the stock X4) or a rigid circuit board (Crazieflie, etc). Either way, I knew I introduced some error here during glue-up; if the angular error was visible without a measuring tool, it was probably significant.
Lesson 5: Minimum wall thickness everywhere might not be the best design philosophy when designing a system that should survive moderate crashes. I obsessed over driving as much weight out of the 3D printed parts as possible; the motor mounts, for example, only have a printed volume of 0.52 cm^3. Based on the two observed structural failures, these bits need to be beefed up substantially for the next design iteration.
So what are the next steps? Well.. I'll probably take a break to work on a few other projects. When I feel motivated to return to the X4, I intend to redesign the frame incorporating the above lessons in mind. Comments/suggestions/critiques welcome!
Trust in Cyanoacrylate
zakqwy • 03/24/2015 at 01:06 • 0 comments
Note to self: design-for-manufacture is NOT the same as design-for-assembly. Especially when DFM translates into "who cares, we're 3D printing it anyway, I'll just watch wall thickness and drain holes and be good to go". Mocking up the X4 frame proved quite difficult; the 4mm spacers that are supposed to hold the CFRP plates AND the arms in place don't like staying still, and the arms are... well, they're round, so they roll off the table a lot.
Space Glue, my previous CFRP weapon of choice, features an unfortunate 3-hour cure time. I couldn't figure out any ways to align the parts that would require less than half a dozen or so glue-ups, so I nixed this idea due to impatience. Fortunately, I located a tube of cyanoacrylate--just the tool for the job!
I've always been leery of Super Glue. I mean.. I always hear it has its place, and maybe my bad experiences had more to due with my tendency to expect more gap-filling properties than actually exist, but the damn stuff never seems to stick to anything but my fingers. But in this case, concerns about strength and joint brittleness (and my fingers) were outweighed by a strong sense of urgency--if I could get the whole mess glued up tonight, I might be flying by bedtime.
I was pretty excited so I didn't take any assembly pictures. Suffice it to say that everything went fairly smoothly; pulltruded CFRP rods need a good abrasive cutter, so I used my carbide tipped hacksaw. Rotating the tube during the cutting process, as recommended by the manufacturer, is key--otherwise you end up with splintery ends. I scored, snapped, and cleaned up the 0.8mm CFRP plate using a utility knife with minimal difficulty.
It's not perfectly square, but I think this will work fine. A few observations:
- I shortened up the pesky battery lead to keep it from getting caught in the prop.
- The motor leads aren't terribly long. I tried to give myself as much leeway as possible by milling a fairly large square hole in the top CFRP plate.
- The leads were still way too short, to the point that even once I'd soldered one of the rear motor leads in place I didn't have enough slack to flip the board into place. Before getting too frustrated, I ordered some fine-gauge wire and a bottle of liquid electrical tape.
Before putting everything away, I pulled the motors out and ran the most important test of all:
Suffice it to say that I'm quite happy with how the eyes turned out.
Double-check your models, kids.
zakqwy • 03/21/2015 at 18:04 • 2 comments

The SLS Nylon bits arrived last week! I sat on 'em for awhile; I've been digging into another project quite a bit lately (more on that later), so I didn't immediately try putting anything together. When I eventually tried assembling one of the motor mounts, I found that the spare motor shell I had lying around didn't want to fit into its hole, despite the CFRP rod fitting snugly with minimal clean-up.
So what's the deal? Time to get out the busted up calipers.
First up--CFRP rod connection:
The ID should match the CFRP rod, which is 4mm stock. Since the wall thickness is 0.7mm (Shapeways' minimum for this material), the OD should be 5.4mm or so. Everything looks good so far.
Next, the pesky motor connection:
Hmm, the OD matches my expectations (7mm motor diameter + 2x wall thickness). The ID, however... is suspiciously 1mm narrow. Time to check the model...
Yup, that's a problem. If you study the assembly picture I posed a few weeks ago, you can even see that the motor model overlaps the ID of this hole since there isn't a defined edge mark:
Time for some sanding. I broke out the Dremel with some sandpaper and various abrasive bits; eventually, I managed to open the motor ID enough to work (getting the part filthy in the process):
During fitment tests I found that the motor was still a bit too snug, which caused the screw portion to distort (the edges of the adjustment slot would cease to be parallel). To fix this without going completely crazy with the sandpaper, I cut a slot below the screw holes to keep them in line:
The process will take some time for all four legs, but I learned a few valuable lessons. Firstly--double-check your CAD models, and then double-check them again. I believe Cubify (like most assembly-capable modeling programs) has some way to check for interferences or at least highlight overlaps; this would have been a valuable step that I ought to investigate in the future.
Secondly--and this bit is quite positive--I learned that I should place more trust in SLS Nylon. Even knowing that it should have similar (if slightly degraded) properties to molded/extruded/formed Nylon, I wasn't prepared for its resiliency. Despite the granular surface texture, it really is strong and flexible. I guess I'm always prepared for textured surfaces to be brittle; this stuff feels like sandstone to the fingers, but it certainly doesn't act like it. As should be expected with a thermoplastic like this, I had better luck removing material with a sharp blade than I did using abrasive methods. Might be a good candidate for a high speed steel mini end mill for the Dremel if I can keep the temperature low.
Thirdly, I'm starting to get my head around the value of personally-owned FDM printers. I've played around with FDM parts in the past (usually ABS) and I haven't been impressed with their strength, density, flexibility, or surface roughness (pretty much anything, I guess). If I'd used an extrusion-type 3D printer to prototype these bits prior to sending them out for production (it's odd to call rapid prototyping "production"), I would have caught the diameter issue early on. Oh well--maybe it's time to add an FDM setup to the list of projects.
In any case, everything worked out in the end:
I think the open back with the wires tucked will protect them adequately, too. I could cut a little plug out of some rubber, or maybe just put a sticker or a piece of tape on the back. Again--not too worried about wire damage at this point:
For now, I'm going to modify the remaining three legs. Then I'll cut the 0.8mm CFRP sheets to size and start fiddling with the other 3D printed bits.
Okay, so these parts are a bit small.
zakqwy • 03/10/2015 at 22:24 • 2 comments

Hardware arrived. Parts are printed and should show up in a day or two. Hand (or part of hand) for scale:
Parts ordered, time for waiting!
zakqwy • 03/04/2015 at 03:17 • 0 comments

Figuring out how to get the various chassis layers to stay together was a bit of a puzzle; I wanted to use tiny socket head cap screws to allow full disassembly, but I also didn't want to drill holes through the arms. Trouble is, the mounting holes on the PCB line up with the arms when everything is centered.
Solution? Press-fit inserts and glued-down SLS parts!
McMaster has a nice selection of 0-80 hardware, so finding these bits wasn't too hard. They also have solid model files of most hardware items, so getting everything into the model was a synch.
I also reduced the size of the motor hardware down to 0-80. Brass nuts for extra awesomeness, and a bit more curve on the loft:
I updated the v03 model to include approximate material densities; based on this info, the finished frame minus props and battery should weigh in around 23g or so. A bit of Internet research suggests that should be a ~20% reduction, so... we'll see how it flies in a few weeks.
Three SLS parts
zakqwy • 02/28/2015 at 17:13 • 2 comments
Getting closer to a print run! Three unique parts at this point:
- Four motor mounts with integrated landing gear
- Four adapters with curvy edges to hold the arms
- Two board spacers with built-in eyes
Pretty straightforward. I bumped the eye model opacity to 100% and temporarily added a hole so they wouldn't look quite so weird:
This stuff all comes together with a few pulltruded CFRP rods, two 0.8mm CFRP plates, the main control board, a tiny rear board spacer, a bit of 2-56 hardware, four motors, a battery, and plenty of Space Glue to look something like this:
Legs got extended to give me a bit more ground clearance; should be close to 10mm now. Wires will run through the arms and connect to the board via a hole in the middle CFRP plate. I haven't added holes at the bottom of the legs to accommodate dabs of hot glue per @Stryker295's recommendation, but this detail will be included prior to order placement.
Thoughts?
Eyes?
zakqwy • 02/27/2015 at 05:18 • 0 comments

I bought a bit of 0.8mm CFRP stock along with my tubing order. I think I may use it as a base plate; the circuit board should be separable from the legs to allow easy motor soldering (easy-ish), but I'm hesitant to build the entire base from SLS nylon. So... time for a spacer of some kind.
Might as well include eyes, right? I tried some 1mm thick parts I had lying around from Shapeways and an ultrabright LED shines right through; the 'pupils' in the model should create a distinct shadow (and I can always sharpie 'em from the inside a bit if needed).
I'll need to orient the print so that the pupils are at the top; with luck, the rings from the print process will show as concentric shadows. In any case, they're interchangable even though only one will use its bolt hole due to circuit board asymmetry. One concern--the antenna is right between the eyes on the board. Might need to pull the CFRP back a bit in that spot to avoid blocking the signal.
MORE RADII!!
zakqwy • 02/26/2015 at 02:28 • 3 comments

I haven't putzed about with the v0.1 base at all, but I think I've got a motor mount design that I like. It's got the split bolt setup for securing the motor (including a cutout for holding a 2-56 nut), and the legs are cool lofted dealies. Lots of radii and impractical features that don't cost anything more to 3D print (the machinist in me is cringing), but look somewhat cool!
Cutouts at the end save a bit of weight and give me access to thread the motor wires through the CFRP tubes. Based on measurements of the original X4 frame (and a bit of trig), I figured out that each motor is actually angled in 1 degree or so; maybe this has a similar effect to toe-in on a car in terms of stability. Either way, I included that feature here--the angle between the CFRP rod and the motor is 89 degrees. Right now the wall thickness of the mounts is 1mm which is the Shapeways minimum.
Think the parts will be strong enough? Any better ideas for feet? I might try to figure out some sort of rubber pad setup to prevent skittering off of tables during slow takeoffs.
Inspirations
zakqwy • 02/25/2015 at 19:41 • 0 comments

This list is far from comprehensive, but these are a few existing projects:
3D printed frame by xm_holger
All-CFRP frame by rz_alex

H-frame by RCaDDiCT!

CFRP design by ridemtb47
3D printed frame by stu121
Got another? Toss a link in the comments and I'll append this list. So far I haven't seen any SLS/CFRP tube combos, but I'm sure they're out there.
Thought--should I make the arm lengths adjustable (somehow)? I initially intended to mirror the stock frame size, but changing it a bit could be fun.
A quick'n'dirty CAD drawing...
zakqwy • 02/20/2015 at 04:03 • 0 comments

I'm planning to 3D print the white bits, so I'll spend some time improving their shape (and generally improving the awesomeness factor of this design). This shows the basic concept; everything is sized roughly right (although the span is a bit wide compared to the actual X4). Battery mount will be (somehow) integrated into the hub.
Don't worry, the eventual model will feature less beefiness on all the nylon bits. I don't want to add too much weight. Oh yeah, and the whole motor mount setscrew deal--you know, the whole reason for this project? I'll add that too.